Upload
vuongmien
View
237
Download
2
Embed Size (px)
Citation preview
2 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Purpose
1. Add value to the data by providing a unified
archive and efficient analysis tools
2. Help using monitoring data to improve
operations, detect and anticipate failures, and
optimize systems
Company Proprietary – Sensitive Information
3 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Big Data Problem?
Company Proprietary – Sensitive Information
• Store all data for any duration?
• Correlation between data sources?
• Further analysis to detect unknown information?
• Learning model to anticipate failures?
AND
THEN …
Systems generate…
data amount
Real-Time processing
Legacy Storage
is everything archived?
How efficiently is data stored
and used?
Company Proprietary – Sensitive Information
4 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Conclusion
Company Proprietary – Sensitive Information
Usual Real-Time Processing
• Virtually keep all data forever… Scale to any size!
• Correlation between data sources!
• Automated analysis for new information?
• Learning model to anticipate failures?
AND
THEN …
Systems generate…
data amount
Big Data
Storage
Company Proprietary – Sensitive Information
5 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Primary Goals
1. Time series monitoring & analytics
2. Low cost system
3. Fast
4. Flexible & Scalable
5. Fault tolerant
6. Incorporates useful analysis features
7. Open to other systems
Company Proprietary – Sensitive Information
6 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
How does it work?
• Time Series Database (KairosDB) • NoSQL Database as storage backend (Apache
Cassandra) • Domain expertise and deep integration
How does it
work?
Architecture
Company Proprietary – Sensitive Information Company Proprietary – Sensitive Information
7 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
How does it work?
ONE single database with a Time Series Web Service frontend
• A Time Series Database frontend (based on KairosDB)
• A NoSQL Database as storage backend (Apache
Cassandra) – We never query from Cassandra directly.
Company Proprietary – Sensitive Information
8 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Large data sets: volume variety and throughput
Monics
TB
PB
GB
MB
Compass
TB
PB
GB
MB
TB
PB
GB
MB
Epoch IPS Neuralstar
TB
PB
GB
MB
Number of metrics: Low
Throughput: High
Storage needs: high
Number of metrics: High
Throughput: Low
Storage needs: Medium
Number of metrics: High
Throughput: High
Storage needs: High
Number of metrics: High
Throughput: High
Storage needs: High
Company Proprietary – Sensitive Information
9 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Solution
Satellite C2 M&C
NMS CSM EPOCH IPS Compass Neuralstar Monics
Other Data Sources
Data Collector agent
Data Collector agent
Data Collector agent
Data Collector agent
Data Integration Frontends
Query, Reporting & Analytics Frontends
Storage (pluggable)
Web UI
External Analytics systems
Company Proprietary – Sensitive Information
10 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
A Typical Skyminer System(s)
Company Proprietary – Sensitive Information
Fault management Data replication Low cost Quick start
Easy administration Scale to any size Best performances
Fault-Tolerant Small Cluster
Using Apache Cassandra DB
Start Small – Scale out at lower cost
when needed
11 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Interoperability Features
1. All features are provided as web services (HTTP / REST + JSON)
2. Open APIs
3. Interoperable data format based on JSON
4. Intuitive Web UI for starting using the system
5. APIs include:
• Data acquisition
• Data querying
• Analysis features (prediction, correlations)
Company Proprietary – Sensitive Information
12 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Existing acquisition connectors
1. Spectrum and Signal Monitoring - Monics data
2. Equipment M&C - Compass data
3. Satellite Control - EPOCH telemetry
4. Enterprise Network Management - Neuralstar
5. Other connectors exist (or can be built)
Company Proprietary – Sensitive Information
TB
PB
GB MB
13 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Toolbox for leveraging
Data value
Company Proprietary – Sensitive Information
14 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Skyminer Exclusive Features
1. Analytics API via ad-hoc queries
• Time Series Query engine
• Horizontal aggregations (Down Sampling) – statistical features calculated over time
• Series combining (vertical aggregations)
• Predictors
• Correlations analysis
2. User Interface
• Web UI with Correlations User Interface
• Integrated Dashboard
Company Proprietary – Sensitive Information
15 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Skyminer Exclusive Features
3. Numerical Analysis
• Fully integrated with R
o The most popular data analysis environment
4. Reporting
• Plugin for BIRT (or its commercial alternative OpenText Analytics)
Company Proprietary – Sensitive Information
16 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Query Engine & aggregations
• Ad-hoc queries and statistics calculation
• Business Intelligence features already implemented (aggregate, drill & pivot)
• Data aggregates: Min, Max, Sum, Average, Count, Rate, Std Deviation…etc
• Multi-level Group-by feature using tags, value, or time
• Filter by tags values
Company Proprietary – Sensitive Information
17 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
• Min • Max • Avg • Sum • Std Dev • Scale • Rate (Derivative) • Least Square • Count • Percentile
Aggregations: Available aggregators
• MinMax • Filter (with predicate) • First • Last • Interpolation • Gaps marker • Alias • Untag/Retag • Time Shift • Formula • Scripted aggregator • Recorder
KairosDB Skyminer
Company Proprietary – Sensitive Information
18 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Time Series Prediction analysis
1. Generic predictive analysis (in the query engine) is being implemented
2. Several predictors : linear (exponential Smoothing, holt, least squares), or dymanic with Dynamic Linear Model (DLM)
Company Proprietary – Sensitive Information
Actual Data Prediction
Below this threshold there is a noticeable impact (failure, degradation)
Predictive analysis indicates when and how-long the system or service will be
affected by the degradation
19 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Correlations Analysis
• Explore and Find correlations within large data sets
• Correlations may indicate relationship in behaviours
• Several analysis methods
• Linear
• DTW (Dynamic Time Warping)
• Other methods can be added as new modules
• Several correlation API and user interfaces
• SEARCH – Find data correlated to a reference
• MATRIX - Explore one-to-one correlations between large number of
series
Company Proprietary – Sensitive Information
20 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Web Interactive Dashboard
1. Integrated Dashboard
2. Most Skyminer analytic features available: aggregators (downsampling, vertical), group-by , filters, and predictors
3. Build a dashboard from various queries in a few mouse clicks from a rich Web user interface
Company Proprietary – Sensitive Information
21 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Reporting
Using BIRT reporting tool plugin
Company Proprietary – Sensitive Information
22 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Specialized Analytics
Company Proprietary – Sensitive Information
Using R Numerical Analysis Environment
23 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Datastore Module
Pluggable DataStore
core
HDF 5
Data Storage back-end can be changed in one line of configuration file => Pick-up the best of the moment for the use case
Company Proprietary – Sensitive Information
24 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Key elements
Company Proprietary – Sensitive Information
25 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Features
• System exists and is operational in production since 2014
• System is: • Fast
• Scalable (1 to N nodes)
• Fault tolerant (1 to N replicas)
• Easy to backup (e.g. Cassandra snapshots files)
• Modular and evolutive
• Open (to change and to other systems) Company Proprietary – Sensitive Information
26 www.integ-europe.com | EUC 2016 : Big Data / SkyMiner |
Conclusion
Skyminer is a simple system, but featuring a rich data processing toolbox for time series
• Unlimited aggregation capabilities
• Analytic features
• Predictions
• Correlations
• Time shifted queries
• Integration with R
Company Proprietary – Sensitive Information